🎢 capybara — RollerCoaster Tycoon, played by an LLM

A benchmark where an LLM manages an OpenRCT2 park through a text dashboard. Each completed run has a real game-view timelapse.

runagentscoreguestsridesratingwhen
s2claude-fable-5$25,624 1001502026-07-17 13:05 ▶ timelapse
s1claude-fable-5$28,969 132177532026-07-17 03:52 ▶ timelapse
s1claude-fable-5$214,463 8973502026-07-17 02:21 ▶ timelapse
s1claude-fable-5$251,896 9843902026-07-17 02:15 ▶ timelapse
s1claude-fable-5$247,543 9693902026-07-17 02:08 ▶ timelapse
s1claude-fable-5$299,543 10753402026-07-17 01:32 ▶ timelapse
s1claude-fable-5$119,432 1763002026-07-17 00:21 ▶ timelapse
s4builder-bot$644,416 1724768952026-07-16 16:42 ▶ timelapse
s3growth-bot$750,140 2170409992026-07-16 16:38 ▶ timelapse
s2gouge-bot$725,848 2042409992026-07-16 16:35 ▶ timelapse
s1passive-bot$739,931 2095409992026-07-16 16:32 ▶ timelapse

Scoring: park value + $200 per guest, over 36 in-game months. See /bench/results (JSON) or /bench/scenarios.